habibst1/Tunisian_Used_Car_Price_Predictor

# 🚗 Car Price Prediction An end-to-end **machine learning application** that predicts the prices of used cars in **Tunisia**. It combines **web scraping, preprocessing, model training, and a Streamlit app** to deliver real-time predictions. --- ## 💡 Use Case This project can be integrated as a **feature for websites that sell used cars in Tunisia**. It allows users to estimate a fair selling or buying price based on the car’s brand, model, year, mileage, and other characteristics. - 🏢 **Car marketplaces** → provide transparent price estimates. - 👤 **Individual sellers** → set competitive prices for their listings. - 👥 **Buyers** → evaluate whether a listed price is reasonable. --- ## ⚙️ How It Works 1. **Data Collection** - Scrapes used car listings from automobile.tn. - Extracts fields: *Brand, Model, Kilométrage, Année, Boîte, Carburant, Puissance fiscale, Prix*. 2. **Data Preprocessing** - Splits **Brand** and **Model**. - Cleans and normalizes categorical and numerical data. - Removes outliers and saves a structured dataset. 3. **Model Training** - Trains multiple ML models (Linear Regression, KNN, Random Forest). - **Random Forest** chosen for best accuracy. - Saves trained model for deployment. 4. **Prediction App (Streamlit)** - User selects **brand and model**. - Inputs car details: mileage, year, fiscal power, gearbox, fuel type. - Model outputs **predicted price in TND**. --- ## 🛠️ Tech Stack - **Python** - **Web Scraping** → `requests`, `BeautifulSoup` - **Data Processing** → `pandas`, `numpy` - **Machine Learning** → `scikit-learn` (Random Forest) - **Visualization** → `matplotlib`, `seaborn` - **Deployment** → `Streamlit` - **Serialization** → `joblib` --- ## 📸 Examples ---